[Talk-ca] duplicate address data

[Gerd Petermann]

Hi Stewart,

I am not sure what you try to say here.  I don't care much about special cases.
My understanding is that the normal scheme in Canada is to have
- odd  numbers on one side and even numbers on the other side of a road
(as in Germany)
- numbers seem to be related to distance, one should not expect to find 20 houses
when an addr:interpolation=both way connects two nodes with 1 and 20
This is also documented here:
http://en.wikipedia.org/wiki/House_numbering#North_America

In Germany I would expect to find 20 houses or a large building with 20 different entries.
as described here:
http://en.wikipedia.org/wiki/House_numbering#Western_and_Southern_Europe

I wanted to point out that the OSM data base for Canada contains a huge amount of 
- useless data like duplicated addr:interpolation ways including nodes from different imports
 which IMHO should be removed ASAP. I don't know how to automate that process,
but I volunteer to help.

- wrong data like 
 * addr:interpolation ways crossing the road
 *  addr:interpolation ways with nodes that refer to a different street 
 *  addr:interpolation ways with nodes that have equal numbers

Do you suggest that programs using the OSM data should tolerate these errors
and try to guess what is meant?
Of course each programmer can do that, but I think the right way is to try to 
have good data in OSM and let programs report the data that is not plausible.

Gerd

Hi Gerd,
> I've tried the latest data for Ontario.

Addressing in Canada is a little bit, um, /special/. And I say that as
someone born in the UK, a country famous for its ultra-baroque addressing.

The important thing I've learned about addresses here is that postal
address ≠ civic address. City addressing is mostly logical, although
with a lot of the municipalities being forcibly amalgamated in 1998,
there are many redundant municipality names/boundaries that /kind of/
matter in addresses. For example, I live in *Toronto*. If you check
Canada Post's address finder, I live in the (former) Township of
*Scarborough*. My postal code, according to Canada Post and almost
everyone on the planet, is M1K 3N_7_. But my civic address, the one for
which I pay property taxes, has a postal code of M1K 3N_8_. But the City
of Toronto sends that tax bill to M1K 3N7, as Canada Post doesn't agree
with the city.

[Some bright spark at a vendor I use decided to rationalize all of the
former municipality addresses into the more modern /Toronto/ — and
immediately broke my pre-authorized credit card payments. Seems that the
card was attached to a Scarborough address, which didn't verify against
Toronto, so payments were stopped.]

Confused yet? Wait until you get to the countryside. There you get
postal addresses which might include a Rural Route number (a mail
delivery route) instead of a street name. There are also County/Township
Route numbers, which are actually street names, but can also have names,
like "Prescott and Russell Road 17", which is County Route 17 on the
border of Prescott & Russell counties. There's yet another address form
in rural areas, which includes the multi-digit 911 number. This is the
emergency services number, and is often given along with the road name.
Canada Post may or may not deliver to a 911 number. And frankly, the
less said about rural postal codes, the better.

What could be usefully done is stripping out redundant address data
where addresses are clearly inside nested administrative boundaries.
There are a lot of addresses that look like this:

    *<tag k="addr:housenumber" v="1045"/>*
    *<tag k="addr:street" v="Pape Avenue"/>*
    <tag k="addr:city" v="East York"/>
    <tag k="addr:province" v="ON"/>
    <tag k="addr:country" v="CA"/>


The last three tags are wholly superfluous, and mean that Nominatim
spits out overly long addresses like “1045, Pape Avenue, Thorncliffe
Park, East York, Toronto, Ontario, M4K 3M6, Canada”.

To one of your interpolation examples:
http://www.openstreetmap.org/node/2009492976 — the end nodes of the
interpolation have "addr:street
<http://wiki.openstreetmap.org/wiki/Key:addr:street?uselang=en-CA> =
County Road 17", but the street itself has "name
<http://wiki.openstreetmap.org/wiki/Key:name?uselang=en-CA> = Prescott
and Russell Road 17". It would be nice if we could interpolate the
nearest parallel(ish) road, rather than needing a name.

cheers,
 Stewart



 		 	   		  
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk-ca/attachments/20150328/3cff6eff/attachment-0001.html>